In a computer system, various letters, combinations of numeric symbols, voices, graphics, images, and so on are generally called data. With the development of technologies such as a social network and an Internet of Things, data keeps increasing at an exponential rate. Increase of a data size also imposes a higher requirement on data retrieval.
Data in a large-size data table is divided into multiple small physical or logical tables according to a policy, and the small tables are stored in different storage units, where each storage unit is a partition. A partition that needs to be scanned is allocated to an idle worker, namely, a processing resource, for example, a processing resource such as a central processing unit (CPU), a thread, a process, or a host; and scanning of a corresponding partition is implemented by these workers.
However, although corresponding required data can be found by using the foregoing retrieval technology, excessively long scanning time of one or more workers may be caused when there is a large data size difference between partitions, thereby leading to excessively long waiting time of a whole retrieval process.